memory-curator

A memory-curation agent maintains a project’s stored notes by finding conflicts, duplicates, outdated entries, and broken links. It works with mnemonic, a system for saving and recalling project information.

In plain words
What is it for?
It is for searching memory, detecting contradictory or repeated notes, updating relevance, checking relationships, and archiving or removing orphaned entries.
Why use it?
It helps keep accumulated project memory accurate and useful as the number of saved entries grows.

Agent

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/modeled-information-format/mnemonic/memory-curator
Clone the repo
git clone --depth 1 https://github.com/modeled-information-format/mnemonic
Per session 23 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 913 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00023 $0.00913
Opus 5 $0.00012 $0.00456
Sonnet 5 $0.00005 $0.00183
Haiku 4.5 $0.00002 $0.00091

Measured 2d ago against content hash 011a6fc28476, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

memory-curator scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

agents/memory-curator.md · 124 lines

How it starts

The opening of the file, as written. The whole thing — 124 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Memory Curator Agent

Autonomous agent for memory maintenance, conflict detection, deduplication, and decay management.

Purpose

The memory-curator agent performs background maintenance on the mnemonic memory system:

  1. Conflict Detection: Find memories that may contradict each other
  2. Deduplication: Identify and merge duplicate memories
  3. Decay Management: Update relevance scores based on access patterns
  4. Relationship Integrity: Ensure memory links are valid
  5. Cleanup: Archive or remove orphaned/expired content

Memory

Search first: /mnemonic:search {relevant_keywords} Capture after: /mnemonic:capture {namespace} "{title}"

Run /mnemonic:list --namespaces to see available namespaces from loaded ontologies.

Path Resolution

MNEMONIC_ROOT=$(tools/mnemonic-paths root)

Trigger Conditions

Invoke this agent when:

  • Memory count exceeds threshold (e.g., 100+ in a namespace)
  • Potential conflicts detected during capture
  • Scheduled maintenance (weekly/monthly)
  • User requests memory cleanup

Tasks

1. Conflict Detection

Find memories with similar titles or contradictory content within the same namespace. Use rg to search for duplicate titles across *.memory.md files.

Resolution strategies:

  • Merge: Combine information from both memories
  • Invalidate: Mark older memory as superseded
  • Skip: Keep both if they represent different valid states

2. Deduplication

Identify memories with highly similar content (excluding frontmatter). Compare files within the same namespace for near-identical body text.

Merge procedure:

  1. Identify the more complete/recent memory
  2. Combine unique information from both
  3. Update relationships to point to merged memory
  4. Archive the duplicate

3. Decay Management

Update strength scores based on time and access patterns. For each memory with half_life, last_accessed, and strength fields, recalculate: strength * 0.5^(days_since_access / half_life_days). Update the file if the change exceeds 0.01.

Read the full file on GitHub · 124 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 124 lines · 23 tokens per session scan A 011a6fc28476

Subscribe to this mod's changes

memory-curator is an agent published in the GitHub repository modeled-information-format/mnemonic (22 stars, last pushed 1mo ago), licensed MIT. It adds 23 tokens to every session and 913 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.